控制权威检索:一种缺失的权威治理知识检索目标
Controlling Authority Retrieval: A Missing Retrieval Objective for Authority-Governed Knowledge
AI总结:
本文提出CAR目标,用于检索当前活跃的权威前沿,与传统相关性检索不同。通过定理和命题分析,证明了TCA的上界,并在多个数据集上验证了方法有效性。
摘要:
在法律中,制药和软件安全的监管制度中,新权威可撤销旧的已确立权威,即使语义上遥远。我们称之为CAR:检索语义锚q的当前活跃权威前沿,即front(cl(A_k(q)))。这不同于寻找最相似的文档:argmax_d s(q, d)。定理4刻画了何时一组R真正覆盖q的活跃权威集,当TCA(R, q)=1时,提供了必要的和充分的条件:前沿包含(front(cl(A_k(q)))包含于R)和无忽略上位者(R外不存在上位文档)。命题2显示TCA@k <= phi(q) * R_anchor(q)在最坏情况下成立,通过对抗性排列论证证明。我们在三个真实世界数据集上进行了评估:安全通告(Dense TCA@5=0.270,两阶段0.975)、SCOTUS推翻对(Dense TCA=0.172,两阶段0.926)和FDA药物记录(Dense TCA=0.064,两阶段0.774)。GPT-4o-mini实验显示,Dense RAG在存在补丁的39%查询中产生明确的'未修补'声明;两阶段将其降至16%。四个基准数据集、领域适配器和单指令评分器已发布在https://github.com/andremir/car-retrieval。
英文摘要:
In law, regulatory regimes for pharmaceuticals and software security, newer authorities can revoke older established ones even when semantically distant. We call this CAR: retrieving the currently active authority frontier for a semantic anchor q, that is, front(cl(A_k(q))). This differs from finding the most similar document by relevance score: argmax_d s(q, d). Theorem 4 characterizes when a set R truly covers the active authority set for q with TCA(R, q)=1, providing conditions necessary and sufficient for any retrieved set R: frontier inclusion (front(cl(A_k(q))) contained in R) and no-ignored-superseder (no superseding document exists in the corpus outside R). Proposition 2 shows that TCA@k <= phi(q) * R_anchor(q) in the worst case over any scope-indexed algorithm, proved by an adversarial permutation argument. We evaluated on three real-world datasets: security advisories (Dense TCA@5=0.270, two-stage 0.975), SCOTUS overruling pairs (Dense TCA=0.172, two-stage 0.926), and FDA drug records (Dense TCA=0.064, two-stage 0.774). A GPT-4o-mini experiment shows Dense RAG produces explicit "not patched" claims for 39% of queries where a patch exists; two-stage cuts this to 16%. Four benchmark datasets, domain adapters, and a single-command scorer are released at https://github.com/andremir/car-retrieval.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。